In dungeons it's honestly hard to test. I'm looking at mobs, threat bars, and my party members locations. On Bosses, looking at boss cast bar, party locations, so can plan ahead. I know when the skill goes off, from my animation of them, not looking at my skill bar. Sounds weird, I know lol.
Ignoring I have Gnashing/Savage/wicked on there twice, only three are keybound, and swap them around when testing, that's how my layout looks, out of the way at the bottom so can pay attention around me, but still there to quickly glance at cooldown timers.
But like I said, tiny hands, so a tight grouping of Keybinds. 1-5, Shift+1-5, QEFCRT, Shift+QEFCRT, Z, X, Alt+1-3, M3 M4 M5, Shift+M3 M4 M5. Dat be a lot of keybinds, and yes I still have Continuation bound since I'm still testing. Thus far I've noticed it heavily depends on player latency, and doing a traceroute to 184.107.107.176(NA Dataserver) results in a 29-37ms ping.
Testing so far, in a vacuum setting of just standing still and hitting a dummy, Macro'd version is 1-2 seconds slower in a 5 minute rotation. But again, this is in a vacuum, perfect scenario of constant damage. It's cost me a total of 19-41 DPS tracked via uh... methods... So I was wrong on DPS increase, just got lucky on crits in that test. But I feel it's not a bad DPS loss, when taking in the averages, for a tank at least.